basquetWi + New ticket
agent-ops OPS-60

cc-context efficiency report - read-only reporting layer on monitor-context.sqlite (week/day/48h)

Backlog normal mcmonitor-context-cc

Sub-tickets

No sub-tickets.
+ Add sub-ticket

Questions

No questions.

Activity

  • system created · 2026-05-22
    wi cli
  • bin-whey-cc assigned · 2026-05-22
    assigned to bin-whey-cc
  • system commented · 2026-05-22
    Spec (PM: nw-venus-cc, filed by nw-whey-cc on whey DB; Elazar confirmed 2026-05-22 00:39). Net-new reader on cc-context-metrics.sqlite (canonical 90-day DB, whey). No new collection - cc-context-monitor.sh write path unchanged. Form: --report mode on cc-context-monitor.sh OR sibling cc-context-report.sh - owner's call. Buckets: last 48h / per-day / per-week. Default metrics per bucket (revisable on Elazar refinement): 1. cache-hit ratio = cum_cache_read / (cum_new_tokens + cum_cache_read) 2. new-tokens-per-turn = cum_new_tokens / cum_turns 3. mean & p90 ctx_tokens vs cap_tokens 4. over_cap sample count progression = deltas of those across buckets. Delivery: llmmsg DM to Elazar now (no msmtp); email once msmtp lands. Scheduled-digest timer = separate child WI assigned nw-whey-cc, file only if/when the report goes ad-hoc->scheduled. Ad-hoc --report needs nothing from nw-whey-cc.
  • wi-cli-whey assigned · 2w ago
    monitor-context-cc / ops
  • wi-cli-whey assigned · 2w ago
    bin-whey-cc -> monitor-context-cc per Elazar ruling 2026-08-01 (monitor-context-cc owns cc-context-monitor/context work). Ownership only; content held pending monitor-context-cc + Elazar fleet-scope confirm.
  • wi-cli-whey note · 2w ago
    PREMISE DEAD: data source cc-context-metrics.sqlite does NOT exist live - trashed on whey (~/.local/share/Trash), absent on venus (monitor-context-cc searched /home/rob depth4). monitor-context-cc's actuator DB is monitor-context.sqlite, a DIFFERENT file, not a rename. Options for owner: re-point a context-efficiency report at monitor-context.sqlite, or close. Do not work as-written.
  • wi-cli-whey note · 2w ago
    Elazar 2026-08-01 (2b): KEEP, re-point the context-efficiency report at monitor-context-cc's live monitor-context.sqlite (venus). Original source cc-context-metrics.sqlite is dead (trashed whey 2026-06-10, never on venus) - do not use it. This makes OPS-60 venus-scoped -> stays with monitor-context-cc (venus-only SSOT per Elazar 1b).
  • wi-cli-venus titleChanged · 2w ago
    cc-context efficiency report - read-only reporting layer on monitor-context.sqlite (week/day/48h)
  • wi-cli-venus commented · 2w ago
    Re-pointed per Elazar ruling 2b (relayed by bin-whey-cc 2026-08-01T08:11): KEEP + re-point at monitor-context.sqlite, stays venus-scoped and mine. Premise correction: cc-context-metrics.sqlite was trashed on whey 2026-06-10T02:28 and never existed on venus, so the original target file is gone, not merely stale. Current state: report.py (~/.local/bin/monitor-context/report.py, registry row 87) was dead since the sh.git move and is now fixed - it reads MONITOR_CONTEXT_DB defaulting to monitor-context.sqlite, with an existence guard so it refuses rather than silently creating an empty DB. Remaining work is ONLY the week/day/48h windows. Not started: report.py is already modified-and-unpushed in sh.git and the push is not yet authorized.
feature
2026-05-22
2w ago